Genetic neural networks for modeling dipole antennas

  • Authors:

  • Petr Šmíd;Zbynek Raida;Zbynek Lukeš

  • Affiliations:

  • Dept. of Radio Engineering, Brno University of Technology, Purkyňova, Brno, Czech Republic;Dept. of Radio Engineering, Brno University of Technology, Purkyňova, Brno, Czech Republic;Dept. of Radio Engineering, Brno University of Technology, Purkyňova, Brno, Czech Republic

  • Venue:
  • AIC'04 Proceedings of the 4th WSEAS International Conference on Applied Informatics and Communications
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

The paper deals with original genetic neural networks for modeling wire dipole antennas. A novel approach to learning artificial neural networks (ANN) by genetic algorithms (GA) is described. The goal is to compare the learning abilities of neural antenna models trained by the GA and models trained by gradient algorithms. Developing the original design method based on genetic models of designed electromagnetic structures is the motivation of this work. Two types of ANN, the recurrent Elman ANN and the feed-forward one, are implemented in MATLAB. Results of training abilities are discussed.